Notes

A news scan found the following data line: Observed on September 11 from 00:06 to 00:12 UTC, a trading system called vectorbt long only with 25 basis points per side was tested. The system was trained using data from 2024 to 2025 and reused from January 1, 2026, to September 10, for a daily matched holdout. The results are shown in a table with columns including look, slippage in basis points, return as a percentage, drawdown as a percentage, win rate, sample size, mean percentage, average win, average loss, half Kelly, and confidence interval low and high. The table has two rows of data: one with a slippage of 10 basis points and another with a slippage of 20 basis points. The returns are mean and equity normalized, showing win, loss, and confidence interval. All 52 saved holdout trades match raw fills plus fees, with an error of less than or equal to 2.84e-14 percent. A neighbor with a period of 10 or 30 was not selected. The next open trade will follow a specific timing, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following data line: Observed on September 11 from 00:06 to 00:12 UTC, the SOL/USD price had a breakout. The trading strategy used is called vectorbt long only, which means it only buys assets, with 25 basis points per side, a measure of the cost of trading. The training data is from 2024 to 2025, and the same data is reused from January 1, 2026, to September 10, which is called the holdout period. The daily matched train is a type of training where the model is trained on daily data. The columns in the data are look, which refers to the trading signal, slippage bp, which is the cost of trading in basis points, return pct, which is the profit or loss as a percentage, drawdown pct, which is the largest decline in value, win rate, which is the percentage of winning trades, sample, which is the number of trades, mean pct, which is the average return, avg win, which is the average profit, avg loss, which is the average loss, half kelly, which is a measure of the optimal bet size, ci low and ci high, which are the lower and upper bounds of the confidence interval, buy hold pct, which is the return of buying and holding the asset, and bars, which is the number of trading periods. The rows in the data are the results of two different trading scenarios, with 20 and 10 as the look values, and the corresponding returns and other metrics. The returns are mean and equity normalized, which means they are adjusted for the risk of the trade. All 52 saved holdout trades match the raw fills and fees, with a difference of less than or equal to 2.84e-14 percent. The neighbor 10/30 was not selected, which refers to a specific trading strategy. The next open cadence, which refers to the timing of the next trade, and no live 15 minute or trailing validation, which refers to the type of validation used, were also noted. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91100 breakout matrix hourly holdout'. This refers to a report about the cryptocurrency Solana, with its price in US dollars, and a specific trading strategy. The report was observed on September 11 from 00:06 to 00:12 UTC. The strategy used a'vectorbt long only' approach, which means it only considered buying the asset, with a 25 basis point, or 0.25%, cost per trade. A basis point is a unit of measurement for interest rates and investment returns. The strategy was trained on data from 2024 and 2025, and then tested on data from January 1, 2026, to September 10, 2026. The results are shown in a table with several columns, including 'look', which refers to the time period considered,'slippage_bp', which is the cost of buying or selling the asset,'return_pct', which is the profit or loss as a percentage, and 'drawdown_pct', which is the largest decline in value. The rows of the table show different combinations of 'look' and'slippage_bp', with values of 10, 20, and 30 for 'look', and 10 and 20 for'slippage_bp'. The results include a'return_pct' of 9.345153% for the first combination, and a 'drawdown_pct' of 27.569157% for the same combination. The report also mentions that all 52 saved holdout trades match the raw fills and fees, with a difference of less than 2.84e-14 percent. The 'neighbor 10/30' was not selected, and the next open trade will be based on a cadence, or schedule, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91100 breakout matrix hourly train'. This refers to a trading strategy for the cryptocurrency Solana, denoted as SOL, against the US dollar, denoted as USD. The strategy was observed on September 11 from 00:06 to 00:12 UTC. The terms'vectorbtlongonly' and '25bp/side' refer to a type of trading strategy and a unit of measurement for trading costs, where 'bp' stands for basis points, a way to express percentages. The 'train2024-25/REUSED2026Jan1-Sep10holdout' indicates the time periods used for training and testing the strategy. The results are presented in a table with columns including 'look','slippage_bp','return_pct', 'drawdown_pct', 'win_rate','sample','mean_pct', 'avg_win', 'avg_loss', 'half_kelly', 'ci_low', 'ci_high', 'buy_hold_pct', and 'bars'. The rows of the table show the results for different combinations of 'look' and'slippage_bp' values. The'returns mean/W/L/CI equity-normalized' refers to the average returns, wins, losses, and confidence intervals, all normalized to the equity. The text also mentions that all 52 saved holdout trades match raw fills plus fees, with a difference of less than or equal to 2.84e-14 percent. The 'Neighbor10/30 not selected' and 'Next-open cadence; no live15min/trailing validation' refer to specific decisions and settings for the trading strategy. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline about a SOL/USD C91100 breakout matrix reconstructed holdout. This is a report from an automated trading fund manager. Observed on September 11 from 00:06 to 00:12 UTC, the system used a method called vectorbt long only, which means it only considers buying, with 25 basis points per side, and trained on data from 2024 to 2025, and reused data from 2026 January 1 to September 10 as a holdout, which is a set of data used to test the system. The system reconstructed the holdout. The report includes several columns of data, including look, which refers to the time period, slippage bp, which is a measure of the difference between expected and actual prices, return pct, which is the percentage return on investment, drawdown pct, which is the percentage decrease in value, win rate, which is the percentage of successful trades, sample, which is the number of trades, mean pct, which is the average percentage return, avg win, which is the average gain, avg loss, which is the average loss, half kelly, which is a measure of the optimal bet size, ci low and ci high, which are measures of confidence, buy hold pct, which is the return if the asset was held, and bars, which is the number of time periods. The rows of data are: [10, 10, 13.537746, 24.589513, 50, 6, 2.670119, 10.784778, 5.44454, 0.123791, -5.267511, 11.59461, -20.818695, 252], [10, 20, 12.183432, 25.040622, 33.333333, 6, 2.464983, 15.939789, 4.27242, 0.077322, -5.456787, 11.371643, -20.818695, 252], [20, 10, -7.586258, 20.455746, 16.666667, 6, -0.917702, 18.777774, 4.856797, -0.024436, -6.581663, 7.325721, -20.818695, 252], [20, 20, -8.688598, 21.010731, 16.666667, 6, -1.115669, 18.540455, 5.046894, -0.030087, -6.768314, 7.111283, -20.818695, 252], [30, 10, -8.194319, 16.712551, 25, 4, -1.475029, 18.777774, 8.225963, -0.039276, -8.658657, 12.073575, -20.818695, 252], [30, 20, -8.925836, 16.962164, 25, 4, -1.671882, 18.540455, 8.409328, -0.045087, -8.841158, 11.849652, -20.818695, 252]. The returns are mean and normalized, and all 52 saved holdout trades match the raw fills and fees, with a difference of less than or equal to 2.84e-14 percent. The neighbor 10/30 was not selected. The next open cadence will be used, and there will be no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91100 breakout matrix reconstructed train'. This refers to a trading strategy for the cryptocurrency SOL in US dollars. Observed on September 11 from 00:06 to 00:12 UTC, the strategy used a'vectorbt long only' approach, which means it only considered buying the asset, with 25 basis points per side, and was trained on data from 2024 to 2025, and reused from January 1, 2026, to September 10, with a holdout period. The reconstructed training data had several columns, including 'look','slippage_bp','return_pct', 'drawdown_pct', 'win_rate','sample','mean_pct', 'avg_win', 'avg_loss', 'half_kelly', 'ci_low', 'ci_high', 'buy_hold_pct', and 'bars'. The rows of data included various combinations of these values, such as 10, 10, -31.457914, 53.121168, 40, 15, -1.3907, 13.49779, 11.31636, -0.051516, -8.55102, 6.901806, -20.794146, and 490. The returns were mean, win, loss, and confidence interval equity-normalized. All 52 saved holdout trades matched raw fills and fees, with a difference of less than or equal to 2.84e-14 percent. The neighbor 10/30 was not selected. The next open cadence was used, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91100 breakout matrix daily holdout'. This refers to a report about the cryptocurrency SOL, which is being traded against the US dollar. The report was observed on September 11 from 00:06 to 00:12 UTC. The trading system used is called vectorbt, which is a type of automated trading system, and it was set to only make long trades, with a 25 basis point, or 0.25%, cost per side. The system was trained on data from 2024 and 2025, and then tested on data from January 1, 2026, to September 10, 2026, which is called the holdout period. The report shows the results of the trading system on the holdout period, with columns including the look, which refers to how much data the system uses to make a decision, slippage, which is the difference between the expected and actual price of a trade, return percentage, drawdown percentage, win rate, sample size, mean percentage return, average win, average loss, half Kelly, which is a measure of the optimal bet size, confidence interval low and high, buy and hold percentage return, and the number of bars, which are periods of time. The rows show the results for different combinations of look and slippage, with values including 10 and 20 for look, and 10 and 20 for slippage. The returns are normalized to the equity, which means they are adjusted to reflect the overall performance of the system. The report also states that all 52 saved holdout trades match the raw fills and fees, with a difference of less than or equal to 2.84e-14 percent. The neighbor 10/30 was not selected, which refers to a specific combination of look and slippage. The next open cadence, which refers to the timing of the next trade, and no live 15 minute or trailing validation, which refers to the system's ability to adapt to changing market conditions, were also noted. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline about a SOL/USD C91100 breakout matrix daily train. The training data was observed on September 11 from 00:06 to 00:12 UTC. The training used a vectorbt long only strategy with 25 basis points per side, and the data was from 2024 to 2025, with a holdout period from January 1, 2026, to September 10. The daily training results are as follows: Columns included look, slippage in basis points, return percentage, drawdown percentage, win rate, sample size, mean percentage, average win, average loss, half Kelly criterion, confidence interval low, confidence interval high, buy and hold percentage, and number of bars. The rows of data are: 10, 10, -31.74928, 53.275836, 40, 15, -1.414632, 13.49779, 11.356247, -0.052402, -8.575156, 6.901806, 586.352958, 493, 10, 20, -33.766398, 54.3188, 40, 15, -1.611606, 13.271021, 11.533358, -0.060719, -8.757823, 6.688216, 586.352958, 493, 20, 10, 21.13546, 31.58847, 44.444444, 9, 3.449974, 19.992952, 9.784408, 0.08628, -7.081952, 15.00619, 586.352958, 493, 20, 20, 18.974523, 31.861571, 44.444444, 9, 3.243281, 19.753205, 9.964659, 0.082095, -7.267603, 14.776407, 586.352958, 493, 30, 10, 13.928635, 29.715413, 42.857143, 7, 3.338385, 22.111861, 10.741723, 0.075489, -8.651304, 17.686589, 586.352958, 493, 30, 20, 12.344743, 30.135859, 42.857143, 7, 3.131914, 21.867881, 10.920061, 0.07161, -8.833819, 17.45145, 586.352958, 493. The returns are mean, win, loss, and confidence interval equity-normalized. All 52 saved holdout trades matched raw fills plus fees, with a difference of less than or equal to 2.84e-14 percent. The neighbor 10/30 was not selected. The next open cadence will be used, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following data line: Observed on September 11 from 00:06 to 00:12 UTC, a trading system had certain settings, including being frozen at 2086 and 2089, using a long-only strategy with 25 basis points per side, and training from 2024 to 2025, then reusing from January 1, 2026, to September 10 for a holdout test. The test had daily matching. The data included several columns such as look, slippage in basis points, return percentage, drawdown percentage, win rate, sample size, mean percentage, average win, average loss, half Kelly criterion, and confidence interval low and high. The rows of data showed two sets of results: one with 20 and 10, and another with 20 and 20, including returns and other metrics. The returns were normalized by equity and showed mean, win, loss, and confidence interval data. All 52 saved holdout trades matched raw fills and fees, with a difference of less than or equal to 2.84e-14 percent. A neighbor selection of 10 and 30 was not chosen. The next open will follow a certain cadence, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following data: Observed on September 11 from 00:06 to 00:12 UTC, a trading system called vectorbt long only with 25 basis points per side was trained from 2024 to 2025 and reused from January 1, 2026, to September 10, with a holdout period. The daily matched train had columns including look, slippage in basis points, return percentage, drawdown percentage, win rate, sample size, mean percentage, average win, average loss, half Kelly, and confidence interval low and high, as well as buy and hold percentage and number of bars. The rows of data were: 20, 10, 31.615468, 33.004847, 40, 15, 2.994078, 20.992348, 9.004768, 0.071314, -4.74328, 10.806615, 30.106603, 729 and 20, 20, 27.725634, 33.339088, 40, 15, 2.788295, 20.750605, 9.186577, 0.067186, -4.933603, 10.585223, 30.106603, 729. The returns were mean and win loss normalized, with all 52 saved holdout trades matching raw fills and fees less than or equal to 2.84e-14 percent. A neighbor 10 30 was not selected, and the next open cadence had no live 15 minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91100 breakout matrix hourly holdout'. This refers to a report about the Ethereum to US Dollar exchange rate. Observed on September 11 from 00:06 to 00:12 UTC, the report used a method called vectorbt, which is a way to analyze and backtest trading strategies, with a long-only approach, meaning it only considers buying, and a 25 basis point, or 0.25%, cost per trade. The report was trained on data from 2024 to 2025 and reused data from 2026 January 1 to September 10 for testing. The results are based on hourly data. The columns in the report include various metrics such as look, which refers to the time period considered, slippage in basis points, return as a percentage, drawdown as a percentage, win rate, sample size, mean percentage return, average win, average loss, half Kelly, which is a measure of the optimal bet size, confidence interval low and high, buy and hold percentage return, and the number of bars, or time periods, considered. The rows in the report show the results for different combinations of parameters. The results are: for a look of 10 and slippage of 10 basis points, the return was 15.336596%, for a look of 10 and slippage of 20 basis points, the return was 13.505891%, for a look of 20 and slippage of 10 basis points, the return was 23.420985%, for a look of 20 and slippage of 20 basis points, the return was 21.948781%, for a look of 30 and slippage of 10 basis points, the return was -7.668646%, and for a look of 30 and slippage of 20 basis points, the return was -8.587361%. The report also mentions that the returns are mean and equity-normalized, and that all 52 saved holdout trades match the raw fills and fees with a difference of less than or equal to 2.84e-14 percentage points. The neighbor 10/30 was not selected, and the next open cadence, which refers to the timing of the next trade, was not validated with live 15-minute or trailing data. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline about an automated trading fund manager's notes. The notes were observed on September 11 from 00:06 to 00:12 UTC. The system used a vectorbt long only strategy with 25 basis points per side, trained from 2024 to 2025 and reused from 2026 January 1 to September 10 as a holdout. This was an hourly train. The columns in the data included look, slippage in basis points, return as a percentage, drawdown as a percentage, win rate, sample, mean percentage, average win, average loss, half Kelly, confidence interval low, confidence interval high, buy and hold percentage, and bars. The rows of data were as follows: [10, 10, 12.52152, 48.602771, 42.857143, 28, 1.372525, 15.3471, 9.108407, 0.044716, -3.673124, 6.79221, 30.106603, 17541], [10, 20, 6.393487, 49.638447, 42.857143, 28, 1.169982, 15.116636, 9.290009, 0.038698, -3.865586, 6.578838, 30.106603, 17541], [20, 10, 61.457476, 37.240047, 52.631579, 19, 3.339892, 14.168035, 8.691378, 0.117867, -2.208573, 9.124278, 30.106603, 17541], [20, 20, 55.437188, 37.677927, 52.631579, 19, 3.133418, 13.939926, 8.873813, 0.11239, -2.403961, 8.906247, 30.106603, 17541], [30, 10, 114.079292, 23.494717, 64.285714, 14, 6.377828, 14.96948, 9.087144, 0.213028, -0.297396, 13.046457, 30.106603, 17541], [30, 20, 108.1682, 23.724004, 64.285714, 14, 6.165285, 14.73977, 9.268788, 0.209138, -0.496603, 12.820589, 30.106603, 17541]. The returns were mean, win, loss, and confidence interval equity-normalized. All 52 saved holdout trades matched raw fills plus fees, which was less than or equal to 2.84e-14 percent per point. The neighbor 10 and 30 were not selected. The next open cadence was used, and there was no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91100 breakout matrix reconstructed holdout'. This refers to a report about the Ethereum to US Dollar exchange rate. The report observed data from September 11 at 00:06 to 00:12 UTC. It used a method called vectorbt long only with 25 basis points per side, trained on data from 2024 to 2025 and reused from January 1, 2026, to September 10, and then reconstructed the holdout data. The report included columns for look, slippage in basis points, return percentage, drawdown percentage, win rate, sample, mean percentage, average win, average loss, half Kelly, and confidence interval low and high, buy and hold percentage, and bars. The rows of data were as follows: [10, 10, -0.805905, 21.137461, 60, 5, 0.125257, 5.896304, 8.531314, 0.010622, -6.408104, 6.658617, -17.85276, 252], [10, 20, -1.792905, 21.60922, 60, 5, -0.074794, 5.684723, 8.714069, -0.006579, -6.595101, 6.445513, -17.85276, 252], [20, 10, -6.40608, 17.452215, 25, 4, -1.403651, 9.496585, 5.037063, -0.073903, -7.803178, 6.297512, -17.85276, 252], [20, 20, -7.151846, 17.864048, 25, 4, -1.600647, 9.27781, 5.2268, -0.086262, -7.987388, 6.085129, -17.85276, 252], [30, 10, -14.24961, 22.098353, 25, 4, -3.439358, 9.496585, 7.751339, -0.181084, -10.075299, 5.217319, -17.85276, 252], [30, 20, -14.932878, 22.487006, 25, 4, -3.632286, 9.27781, 7.935652, -0.195751, -10.254969, 5.007094, -17.85276, 252]. The returns are mean, win, loss, and confidence interval equity-normalized. All 52 saved holdout trades match raw fills plus fees less than or equal to 2.84e-14 percent. The neighbor 10/30 was not selected. The next open cadence was used, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the data line for the ETH/USD C91100 breakout matrix reconstructed train. The data was observed on September 11 from 00:06 to 00:12 UTC. The system used a vectorbt long-only strategy with 25 basis points per side, trained from 2024 to 2025 and reused from 2026 January 1 to September 10 as a holdout. The reconstructed train has several columns including look, slippage in basis points, return percentage, drawdown percentage, win rate, sample size, mean percentage, average win, average loss, half Kelly, and confidence interval. The rows of data are as follows: [10, 10, 61.295947, 38.327629, 40, 20, 3.875261, 23.612019, 9.282577, 0.082061, -3.685228, 12.240373, 30.106603, 729], [10, 20, 54.971428, 38.757921, 40, 20, 3.667718, 23.365041, 9.463831, 0.078487, -3.877665, 12.016116, 30.106603, 729], [20, 10, 31.615468, 33.004847, 40, 15, 2.994078, 20.992348, 9.004768, 0.071314, -4.74328, 10.806615, 30.106603, 729], [20, 20, 27.725634, 33.339088, 40, 15, 2.788295, 20.750605, 9.186577, 0.067186, -4.933603, 10.585223, 30.106603, 729], [30, 10, 109.964557, 15.037477, 54.545455, 11, 8.063495, 21.308857, 7.830939, 0.189205, -1.056922, 16.986546, 30.106603, 729], [30, 20, 105.395767, 15.122525, 54.545455, 11, 7.847583, 21.066481, 8.015094, 0.186258, -1.25461, 16.752806, 30.106603, 729]. The returns are mean and equity-normalized, showing win, loss, and confidence interval. All 52 saved holdout trades match raw fills and fees, with an error of less than or equal to 2.84e-14 percent. The neighbor 10/30 was not selected. The next open cadence will be used, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91100 breakout matrix daily holdout'. This is a report from an automated trading fund manager. Observed on September 11 from 00:06 to 00:12 UTC, the system used a method called vectorbt, which is a way to analyze and make decisions about trades, with a long-only strategy, meaning it only buys and never sells short, and a risk management setting of 25 basis points per side, which is a measure of the risk taken. The system was trained on data from 2024 to 2025 and reused from January 1, 2026, to September 10, and it is now in a daily holdout period, meaning it is not using new data to make decisions. The report includes several columns of data, including the look, which refers to the time period being analyzed, slippage in basis points, return as a percentage, drawdown as a percentage, win rate, sample size, mean percentage return, average win, average loss, half Kelly, which is a measure of the optimal bet size, and confidence interval low and high. The rows of data show the results of different scenarios, with 10 and 20 referring to the look period, and the numbers in the rows representing the results of the different scenarios. The returns are mean and equity-normalized, meaning they are adjusted to reflect the risk taken, and all 52 saved holdout trades match the raw fills and fees, with a difference of less than 2.84e-14 percent. The neighbor 10/30 was not selected, and the next open cadence, which refers to the timing of the next trade, will not include live 15-minute or trailing validation, which are methods of evaluating the performance of the system. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91100 breakout matrix daily train'. Observed on September 11 from 00:06 to 00:12 UTC, the system was frozen, using a vector backtest with long only positions, 25 basis points per side, and training data from 2024 to 2025, with a reused 2026 dataset from January 1 to September 10 as a holdout. The daily training results are as follows: the columns represent different metrics such as look, slippage in basis points, return percentage, drawdown percentage, win rate, sample size, mean percentage, average win, average loss, half Kelly criterion, confidence interval low, confidence interval high, buy and hold percentage, and the number of bars. The rows represent different scenarios with the following results: [10, 10, 19.129186, 46.817438, 36.363636, 22, 2.228971, 23.612019, 9.989913, 0.0472, -5.039284, 9.892798, 30.106603, 731], [10, 20, 14.001134, 47.294013, 36.363636, 22, 2.024717, 23.365041, 10.169754, 0.043328, -5.229016, 9.673231, 30.106603, 731], [20, 10, 16.773358, 33.004847, 37.5, 16, 2.102144, 20.992348, 9.231979, 0.050069, -5.283678, 9.836916, 30.106603, 731], [20, 20, 13.095757, 33.339088, 37.5, 16, 1.898143, 20.750605, 9.413334, 0.045737, -5.472921, 9.617462, 30.106603, 731], [30, 10, 86.28712, 15.06911, 50, 12, 6.451797, 21.308857, 8.405262, 0.151388, -2.213318, 15.266054, 30.106603, 731], [30, 20, 81.869443, 15.154126, 50, 12, 6.239106, 21.066481, 8.588269, 0.148081, -2.408696, 15.035752, 30.106603, 731]. The returns are mean, win, loss, and confidence interval equity-normalized. All 52 saved holdout trades match raw fills plus fees, with an error of less than or equal to 2.84e-14 percent. The neighbor 10/30 was not selected. The next open cadence will be used, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following data line: Observed on September 11 from 00:06 to 00:12 UTC, a trading system had certain settings, including a type of trading strategy called vectorbt long only, with 25 basis points per side, and training data from 2024 to 2025, and also from January 1, 2026, to September 10, which was not used for training, called holdout. The system looked at several things, including the look, which is the time period considered, slippage in basis points, which is the difference between expected and actual trade prices, return as a percentage, drawdown as a percentage, which is the decrease in value from a peak, win rate, sample size, mean return as a percentage, average win, average loss, half Kelly, which is a betting strategy, confidence interval low and high, buy and hold return as a percentage, and the number of bars, which are units of time. The system found two possible sets of results: one with 20 and 10, and a return of 4.831708 percent, and another with 20 and 20, and a return of 3.788612 percent. The system also found that all 52 saved trades in the holdout period matched the actual trades, including fees, with a very small difference of 2.84e-14 percent. The system did not select a neighbor with a distance of 10 or 30. The next trade will be done at the next open, and there will be no live validation every 15 minutes or using a trailing method. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline is not available, but a data line was found. The data is about a trading strategy for Bitcoin, observed on September 11 from 00:06 to 00:12 UTC. The strategy used a long-only approach, with 25 basis points per side, and was trained on data from 2024 to 2025, and then reused from January 1, 2026, to September 10, 2026, as a holdout. The daily matched train had several columns, including look, slippage in basis points, return as a percentage, drawdown as a percentage, win rate, sample, mean percentage, average win, average loss, half Kelly, and confidence interval low and high. The rows of data showed two sets of results, with 20 and 10 basis points of slippage, and 20 and 20 basis points of slippage, with returns of 0.56998 and -2.791931, respectively. The strategy's returns were mean and equity-normalized, and all 52 saved holdout trades matched raw fills and fees, with an error of less than or equal to 2.84e-14 percent. A neighbor with a period of 10 or 30 was not selected, and the next open cadence was used, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91100 breakout matrix hourly holdout'. This refers to a report about Bitcoin trading, where a computer program made decisions to buy or sell Bitcoin. The report was observed on September 11 from 00:06 to 00:12 UTC. The program used a method called vectorbt, which is a way to analyze and make decisions about trading, and it only considered buying, not selling. The program was trained using data from 2024 and 2025, and then it was tested on data from 2026, from January 1 to September 10, to see how well it would work. The results are shown in a table with many columns, including the return as a percentage, the drawdown as a percentage, and the win rate, which is the percentage of trades that were successful. The table has five rows of numbers, each representing a different set of conditions, with numbers such as 10, 20, 13.659676, and 50. The report also mentions that the returns were normalized, meaning they were adjusted to make them comparable, and that all 52 trades that were saved and held out for testing matched the actual trades, including fees, with a very small difference of less than 2.84e-14 percent. The report also mentions that a 'neighbor' with a value of 10 or 30 was not selected, and that the next step is to open a trade at a certain time, without validating it first. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline about a breakout matrix for BTC/USD on September 11 from 00:06 to 00:12 UTC. The system used a method called vectorbt, which is a way to analyze trading data, and it was set to only make long trades, with 25 basis points per side, and it was trained on data from 2024 to 2025, and reused data from 2026 January 1 to September 10 as a holdout, which means it was not used for training. The system was trained on hourly data. The results had several columns, including look, slippage in basis points, return as a percentage, drawdown as a percentage, win rate, sample size, mean return as a percentage, average win, average loss, half Kelly, which is a way to measure risk, and confidence interval low and high, buy and hold return as a percentage, and the number of bars, which are periods of time. The rows of data were as follows: [10, 10, -0.004706, 43.835941, 34.615385, 26, 0.395394, 11.112296, 5.27826, 0.017791, -2.960033, 4.090433, 106.954861, 17541], [10, 20, -5.071592, 45.332167, 34.615385, 26, 0.194803, 10.890293, 5.467515, 0.008944, -3.15392, 3.88246, 106.954861, 17541], [20, 10, 18.04607, 28.481577, 36.842105, 19, 1.29518, 12.054157, 4.98089, 0.053723, -2.748958, 5.573684, 106.954861, 17541], [20, 20, 13.644469, 29.334671, 36.842105, 19, 1.092791, 11.830272, 5.170739, 0.046186, -2.943266, 5.362747, 106.954861, 17541], [30, 10, 18.080262, 25.505467, 43.75, 16, 1.497226, 11.035447, 5.921391, 0.067837, -3.164637, 6.371631, 106.954861, 17541], [30, 20, 14.361502, 26.099047, 43.75, 16, 1.294434, 10.813597, 6.10936, 0.059852, -3.358114, 6.1591, 106.954861, 17541]. The returns were mean and normalized by equity, and all 52 saved holdout trades matched the raw fills and fees, with a difference of less than or equal to 2.84e-14 percent. The neighbor 10 and 30 were not selected. The next open cadence was used, and there was no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91100 breakout matrix reconstructed holdout'. This refers to a report about a trading strategy for Bitcoin, which is a digital currency, and the US dollar. The report says that on September 11, from 00:06 to 00:12 UTC, a system called vectorbt, which is a tool for backtesting trading strategies, was used to test a long-only strategy, meaning it only buys and never sells short, with a 25 basis point, or 0.25%, cost per trade. The system was trained on data from 2024 to 2025 and reused for 2026 from January 1 to September 10. The results of the test are shown in a table with columns for different metrics, such as return percentage, drawdown percentage, and win rate, which is the percentage of trades that were profitable. The rows of the table show the results for different combinations of parameters, including the look, which refers to the amount of data used to make a trading decision, and slippage, which is the difference between the expected price of a trade and the actual price. The results include the mean return percentage, average win, average loss, and other metrics. The report says that all 52 trades in the test match the raw data, including fees, with a very small difference of less than 2.84e-14 percent. The report also says that a neighbor, which refers to a similar trading strategy, with parameters 10 and 30, was not selected, and that the next open trade will be made according to a regular schedule, without validation every 15 minutes or when the price is trailing, meaning moving in the opposite direction of the trade. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline about a BTC/USD C91100 breakout matrix reconstructed train. Observed on September 11 from 00:06 to 00:12 UTC, the system was frozen, using 2086 and 2089 data points, and a long-only strategy with 25 basis points per side, trained from 2024 to 2025 and reused from 2026 January 1 to September 10 as a holdout. The reconstructed training data has several columns, including look, slippage in basis points, return percentage, drawdown percentage, win rate, sample size, mean percentage, average win, average loss, half Kelly criterion, and confidence interval low and high, buy and hold percentage, and number of bars. The rows of data are as follows: [10, 10, -34.267669, 56.309969, 32, 25, -1.123916, 11.820645, 7.215474, -0.04754, -5.019263, 3.458366, 106.954861, 729], [10, 20, -37.47348, 58.232507, 32, 25, -1.321471, 11.597226, 7.400858, -0.056974, -5.209035, 3.251655, 106.954861, 729], [20, 10, 0.56998, 31.090522, 41.176471, 17, 0.59803, 11.264459, 6.868471, 0.026545, -4.387682, 5.964294, 106.954861, 729], [20, 20, -2.791931, 33.060304, 41.176471, 17, 0.397034, 11.042152, 7.054548, 0.017978, -4.578716, 5.752577, 106.954861, 729], [30, 10, -1.810727, 33.086388, 38.461538, 13, 0.285013, 9.963841, 5.764255, 0.014302, -4.524629, 5.705514, 106.954861, 729], [30, 20, -4.330752, 33.619562, 38.461538, 13, 0.084643, 9.744133, 5.952538, 0.004343, -4.71539, 5.494314, 106.954861, 729]. The returns are mean, win, loss, and confidence interval equity-normalized, and all 52 saved holdout trades match raw fills plus fees, with a difference of less than or equal to 2.84e-14 percent. The neighbor 10 and 30 were not selected, and the next open cadence will be used, with no live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91100 breakout matrix daily holdout'. This is a report from an automated trading fund manager. The report is about a trading strategy for Bitcoin priced in US dollars. It was observed on September 11 from 00:06 to 00:12 UTC. The strategy used a long-only approach, which means it only bought assets, with a 25 basis point cost per trade. A basis point is a unit of measurement, one hundredth of one percent. The strategy was trained on data from 2024 and 2025, and then tested on data from January 1, 2026, to September 10, 2026. The results are shown in a table with several columns, including the look, which refers to the time period, slippage, which is the difference between expected and actual trade prices, return percentage, and drawdown percentage, which is the decrease in value from a peak. The rows in the table represent different combinations of look and slippage. The results show that the strategy had a negative return percentage and a high drawdown percentage in all cases. The mean return percentage was also negative. The strategy's performance was compared to a buy-and-hold approach, which means buying an asset and holding it for a long time. The report also mentions that the strategy's returns were normalized, which means they were adjusted to account for the equity, or value, of the assets. The report concludes that a neighboring strategy with a look of 10 or 30 was not selected, and the next trade will be made at the next open, without validation every 15 minutes or when the price is trailing. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91100 breakout matrix daily train'. This is a report from a trading fund manager. The report is about a trading strategy for Bitcoin, observed on September 11 from 00:06 to 00:12 UTC. The strategy is called'vectorbt long only' with 25 basis points per side, which means a small amount of money is used for each trade. The strategy was trained using data from 2024 to 2025 and reused from January 1, 2026, to September 10, 2026, as a test period. The report shows the results of the strategy in a table with columns such as 'look','slippage bp','return pct', and 'drawdown pct'. Slippage refers to the difference between expected and actual trade prices. The rows in the table show different combinations of parameters, with 10, 20, and 30 values for 'look' and 10 and 20 values for'slippage bp'. The results include return percentages, drawdown percentages, and win rates. The report also mentions that the returns are normalized, meaning they are adjusted to account for the risk taken. All 52 saved trades match the raw trade data, with very small differences of less than 2.84e-14 percent. The report concludes that the 'neighbor 10/30' strategy was not selected and the next open trade will be made without live 15-minute or trailing validation. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91100 10000-path circular block bootstrap risk scenario'. This refers to a risk assessment for the SOL/USD currency pair. The assessment uses 10,000 paths, a 20-day horizon, and 3-day blocks. It is based on 253 daily close returns and has a closing price of 98.6025. The average true range over 14 days is 4.975069381535714. The stop loss percentage is 10.091162762679872 and the target percentage is 30.27348828803964. The stop loss percentage for the first day is 42.68 and the target percentage for the first day is 4.109999999999999. The percentage of scenarios where neither the stop loss nor the target is reached is 53.21. The terminal percentage at the 5th percentile is -24.072090275644573, at the 50th percentile is -1.2337437517659422, and at the 95th percentile is 25.23982856754894. The account loss at the 5th percentile, assuming a notional value of 212 and holding for 20 days without stops, is -6.018022568911143. This scenario does not account for fees, intraday price movements, or 15-minute execution times. It is an unconditional scenario and does not consider the probability of a breakout or validated exit. The account loss figure assumes the position is held for 20 days without stops and does not account for the possibility of a gap-through-stop Value at Risk. https://runway.adityaperswal.workers.dev/data/crypto-bars